Many potential use cases for machine learning in chemistry and materials science suffer from small dataset sizes, which demands special care for the model design in order to deliver reliable predictions. Hence, feature selection as the key determinant for dataset design is essential here. We propose a practical and efficient feature filter strategy to determine the best input feature candidates. We illustrate this strategy for the prediction of adsorption energies based on a public dataset and sublimation enthalpies using an in-house training dataset. The input of adsorption energies reduces the feature space from 12 dimensions to two and still delivers accurate results. For the sublimation enthalpies, three input configurations are filtered from 14 possible configurations with different dimensions for further productive predictions as being most relevant by using our feature filter strategy. The best extreme gradient boosting regression model possesses a good performance and is evaluated from statistical and theoretical perspectives, reaching a level of accuracy comparable to density functional theory computations and allowing for physical interpretations of the predictions. Overall, the results indicate that the feature filter strategy can help interdisciplinary scientists without rich professional AI knowledge and limited computational resources to establish a reliable small training dataset first, which may make the final machine learning model training easier and more accurate, avoiding time-consuming hyperparameter explorations and improper feature selection.

We develop a theoretical model to predict the sublimation vapor pressure of pure substances. Moreover, we present a simple monoatomic molecule approximation, which reduces the complexity of the vapor pressure expression for polyatomic gaseous molecules at a convincing level of accuracy, with deviations of the Arrhenius prefactor for $\mathrm{NaCl}$ and $\mathrm{NaF}$ being 5.02% and 7.08%, respectively. The physical model is based on ab initio calculations, statistical mechanics, and thermodynamics. We illustrate the approach for $\mathrm{Ni}$, $\mathrm{Cr}$, Cu (metallic bond), $\mathrm{NaCl}$, $\mathrm{NaF}$, ${\mathrm{ZrO}}_2$ (ionic bond) and ${\mathrm{SiO}}_2$ (covalent bond). The results are compared against thermodynamic databases, which show high accuracy of our theoretical predictions, and the deviations of the predicted sublimation enthalpy are typically below 10%, for $\mathrm{Cu}$ even only 0.1%. Furthermore, the partial pressures caused by gas phase reactions are also explored, showing good agreement with experimental results.

This study validates a novel method for simultaneous durability testing of multiple symmetrical cells with Ni/ceria fuel electrodes. The investigation demonstrates that gas diffusion losses in the multi-cell test setup utilizing the active driven gas layer concept have a smaller impact on cell performance compared to the single-cell setup. This method proves effective for testing multiple identical cells, offering a time and cost-efficient approach. However, testing symmetrical cells with different fuel electrodes reveals an unexpected pseudo-inductive loop in the impedance spectra, observable only in the multi-cell setup. Analysis of impedance spectra and relaxation times indicates that differences in electrode polarization resistance result in varying gas compositions at the two electrodes of a cell, disrupting symmetry and superimposing an additional Nernst voltage term. To understand this, an electrical equivalent circuit model with the Nernst voltage term was developed to reproduce the observed behavior. Comparison of experimental and simulation results substantiates the model and elucidates the mechanisms. The findings indicate that testing multiple cells with the active driven gas layer concept is applicable only for identical cells and not for those with different electrodes.